An interactive environment for motor unit potential classification using certainty-based classifiers

https://doi.org/10.1016/j.simpat.2008.06.008Get rights and content

Abstract

An interactive environment for performing the motor unit potential (MUP) classification tasks required for electromyographic (EMG) signal decomposition is described and developed utilizing the MATLAB high-level programming language and its interactive environment. Certainty-based classification approach had been employed for the classification task in which the assignment criterion used for MUPs is based on a combination of MUP shapes and motor unit firing pattern information. The environment software package consists of several graphical user interfaces used to detect individual MUP waveforms from a raw EMG signal, extract relevant features, and classify the MUPs into motor unit potential trains (MUPTs) using certainty-based classifiers. The development of the proposed software package is useful at enhancing the analysis quality and providing a systematic approach to the EMG signal decomposition process.

Introduction

Interaction (dialogue) with a computer is a style of control and interactive systems exhibit that style [18]. Our goal in this paper is to present an interactive system that works as an interactive environment for the classification task required for electromyographic (EMG) signal decomposition through developing a MATLAB interactive software package. The developed system is used for classifying motor unit potential (MUP) waveforms using certainty-based classifiers [13], [15], [17], [21]. The results of the certainty-based classifiers reported in [15] have been generated using the developed interactive environment presented in this paper. The purpose of interaction in the developed system is to facilitate the use of a computer for MUP classification and to enhance the user’s power to accomplish this task.

The MUP classification interactive environment consists of tasks involving the user and the interactive system. These tasks can take place in a sequential fashion. Tasks are hierarchical and at each hierarchical level feedback for previous actions is required. These tasks are performed as a sequence of messages between the user and the computer running the interactive system and forming an interactive dialogue. The user’s messages are transmitted via a set of control objects and input devices while the computer’s messages are transmitted via a set of displays.

The inputs and outputs of the interactive MUP classification system are modelled as languages. The language goals are to provide methods for describing the user interface and the programs that model the interface in the developed interactive system. In the interactive world, two interfaces to the computer are distinguished. The first between the user and the computer is called the user interface. The second between the programmer of the system and the computer is called the program interface. The user interface provides the means to communicate with the computer by using the dialogue language. The dialogue language is handled by its counterpart on the programmer side: the programming language.

The developed software package provides graphical user interfaces (GUIs) to detect individual MUP waveforms from a raw EMG signal, extract relevant features, classify the detected MUPs into motor unit potential trains (MUPTs) using different versions of certainty-based classifiers: the certainty classifier (CC) [13], [21], and the adaptive certainty classifier (ACC) [15], [17]. As a classification problem, the MUP waveforms represent the patterns to be recognized and the MUPTs represent the classes into which the MUP waveforms are to be grouped. The criterion for grouping MUPs is based on a combination of MUP shape and motor unit (MU) firing pattern information.

Section snippets

EMG signal analysis

An electromyographic (EMG) signal is the recording of the electrical activity associated with muscle contraction, and reflects the electrical depolarization of excitable muscle fibre membranes that create electrical signals called muscle fibre potentials (MFPs). It can be detected by applying surface electrodes to the skin overlying a muscle. Surface EMG signals are easy to measure and noninvasive. But, for diagnostic purposes it is useful to employ concentric needle electrodes that are

Certainty-based classifiers

A certainty-based classifier classifies a candidate MUP to the MUPT that produces the greatest estimated certainty, provided this maximal certainty is above a minimum certainty threshold.

Structure of the interactive environment

A program architecture for the MUP classification interactive environment is needed for enabling the user to control interactively to a certain extent the flow of activities during program execution. Normally, the user can exercise influence on a program’s control flow only at points where the program is ready for it. The user interface of the program represents the means by which the program allows the user to influence the flow of control during execution. The user interface consists of

Feature extraction

The first task in EMG signal decomposition is the segmentation and MUP detection task. It is concerned with locating the main positive peaks or spikes found in an EMG signal. The detected spikes or MUPs should have rapid rising edges, which indicates that the electrode is close to active muscle fibres. MUs that were active during signal acquisition generate these MUPs. Conversely, MUPs that have slow rising edges and small amplitudes were generated from MUs with fibres that were far away from

Supervised classification of MUPs

The task of supervised classification during the process of EMG signal decomposition is involved with the discrimination of the activation patterns of individual MUs, active during contraction, into distinguishable MUPTs. Therefore, MUPs most likely belong to the same MUPT if their shapes are closely similar and if their IDI intervals are consistent with the discharge pattern of the considered MU. This means that two kinds of information, the MUP shapes and the times of occurrences of MUPs,

Application in decomposing an EMG signal

In this section we will present the results of decomposing a test EMG signal. The test EMG signal is a simulated signal generated by the EMG simulator which is integrated within the developed system. Fig. 12, Fig. 13 show some details concerning the tested signal and Fig. 14 shows the ACC classifier performance in terms of the performance indices and the predicted confusion matrix when classifying the test signal. The test EMG signal used has eight MUPTs and was simulated to have an intensity

Conclusions

In this paper, we presented and constructed a MATLAB interactive software package consisting of a set of integrated graphical user interfaces (GUIs) to efficiently help users work in an interactive environment for MUP classification during EMG signal decomposition. The package relieves users from the boring and painful efforts associated when dealing with the extensive textual information associated with the MUP classification task; provides users with a full control of MUP classification

References (26)

  • A.B. Geva

    Application of fuzzy clustering to biomedical signal processing and dynamic system identification

  • A. Hamilton-Wright et al.

    Physiologically based simulation of clinical EMG signals

    IEEE Transactions on Biomedical Engineering

    (2005)
  • S. Kanoun, A new digital signal processing technique for the estimation of motor unit action potential templates,...

    • Cited by (3)

    View full text
    Copyright © 2008 Elsevier B.V. All rights reserved.