KNOWLEDGE-BASED SIGNAL PROCESSING OF MYOELECTRIC SIGNALS

Abstract

This dissertation describes an automatic signal processing method for extracting and characterising motor unit action potential (MUAP's) from the electromyographic interference pattern for clinical as well as research•aplication using off-line system and also describes how these features of MUAP's will be useful in building a knowledge-based system (KBS). The EMG signal is weighted sum of the electrical activity of the recruited motor units (MU) and a motor unit action potential (MUAP) is the spatial and temporal summation of all muscle fibres. in MU. Its analysis, in addition to establish -normative data bases, will improve communication about and description of both normal and abnormal activities and will permit more accurate serial comparision of EMG signals. To start with, an attempt has been made to decompose MUAP's from EMG signal in the time domain. For this, paper recorded signal is digitized and data are stored in a IBM comptible PC. The stored data are digitally filtered to transform the sharp rising edges of the MUAP's into narrow spikes better suited for detection and classification. The spikes that exceed a certain detection threshold are classified and equal spikes are averaged. Finally each spike is characterized by its amplitude, duration, slope, number of phases etc.