Engineering 5, 6th Floor
519-888-4567, Extension 32600
Office: EC4 2023
Phone: (519) 888-4567 ext. 32982
Fax: (519) 746-4791
- BASc - University of Waterloo
- MEng - McMaster University
- PhD - McMaster University
Electrophysiological Characterization of Neuromuscular Disorders:
Muscles are functionally composed of motor units which are groups of muscle fibres simultaneously innervated by a single alpha-motorneuron. Muscle contractions are produced by the controlled repetitive activation of groups of MUs. Each activation of a motor unit produces a region of transmembrane current on each fibre of the motor unit, which propagates along the contacting fibre.
Via volume conduction, a changing potential field, which is the temporal and spatial summation of contributions from all active muscle fibres, is created throughout the interstitial fluid surrounding the muscle. Suitable electrodes can detect this electrical activity. The characteristics of the detected electromyographic (EMG) signals and, therefore, the type and amount of information which is available through their analysis are dependent on the electrodes used, the level and type of contraction and the specific muscle structure.
Disorders of muscle and nerve change the structure and organization of the muscles involved. Characteristics of clinical EMG signals suitably detected from affected muscles reflect the degree and type of disorder present. Physicians currently, subjectively and qualitatively, assess EMG signal characteristics to support diagnosis, treatment and management of specific disorders. This assessment is highly dependent on physician skill and experience. Furthermore, assessment of the severity of involvement and the sensitivity with which progressive changes can be tracked is limited. Quantitative electromyography (QEMG) involves the detection and analysis of EMG signals to calculate statistics for neuromuscular characterization. QEMG can improve the specificity and sensitivity of neuromuscular assessments and track longitudinal changes associated with specific treatment or management regimes. However, a method for completing the crucial step of interpreting QEMG data to produce a neuromuscular characterization is currently needed.
The current major objectives of our ongoing research program are as follows:
1. To develop and evaluate methods that can assist in the interpretation of QEMG results.
2. Introduce new quantitative statistics that more effectively reflect neuromuscular structure and physiology.
Based on clinically viable EMG signal decomposition systems that provide QEMG data containing information regarding the structural, organizational and operational state of the muscles under study, novel statistically-based pattern discovery techniques for extracting underlying information and facilitating interpretation of QEMG results by creating a neuromuscular characterization are being developed and evaluated. Characterizations are to be presented in linguistic terms, with their rationale easily understood and their statistical basis able to be examined. In addition, valid measures of the confidence in a suggested interpretation are being investigated. The developed system by facilitating interpretation of QEMG data will greatly increase its usefulness and ultimately its clinical use.
EMG Signal Decomposition:
Motor units are repetitively active during sustained voluntary contraction. Each activation of a motor unit results in contraction of its fibres and associated electrical activity that, when detected, is called a motor unit potential (MUP). Repeated activation of a motor unit produces a motor unit potential train (MUPT). EMG signals are composed of the summation of the MUPTs of all active motor units. The process of EMG signal decomposition resolves a composite signal into its constituent MUPTs. EMG signal decomposition involves the application and development of pattern recognition techniques and associated signal processing and machine intelligence algorithms.
New methods to decompose an acquired composite needle-detected EMG signal into its constituent MUPTs are being developed. The resulting MUPTs represent the activity of individual motor units and can therefore provide firing pattern and morphological information regarding the population of motor units active during signal detection with fibres suitably close to the electrode. The number of motor units studied during EMG signal decomposition typically ranges from 4 to 12 for slight to moderate levels of contraction. The current EMG signal decomposition system is composed of a series of algorithms. MUPs are detected, clustered (using both shape and firing pattern information), and classified (using a multi-pass supervised classification algorithm that uses relative distance measures and firing pattern information). Statistically based algorithms then analyze the firing patterns of the resulting MUPTs in search of temporal interdependencies to determine whether trains should be merged or disregarded. The decomposition system has been evaluated using a set of signals obtained from normal muscle. Preliminary results suggest that the decompositions obtained are sufficiently accurate, complete and robust to be clinically useful. However, further evaluation of the system when used with signals obtained from subjects with various neuromuscular disorders is required.
EMG signals are detected from skeletal muscle during voluntary contraction or following stimulation of muscle or nerve. Characteristics of the electrodes, the contraction or stimulus and the underlying neuromuscular system determine the characteristics of an EMG signal. Disorders of muscle and nerve change the structure and organization of the muscles involved. Therefore, if a standard detection protocol is used, EMG signal characteristics can help determine the health of an underlying neuromuscular system. QEMG attempts to extract clinical neuromuscular information from suitably detected EMG signals. Studying individual MUPs and MUPTs can provide valuable morphological and temporal information regarding a muscle and its motor neurons, which can be used in the diagnosis of neuromuscular disorders and in the study of neuromuscular control mechanisms. QEMG results are becoming more widely used and more heavily weighted by clinical neurophysiologists when making clinical decisions related to characterizing and diagnosing neuromuscular disorders. In collaboration with clinical neurophysiologists, significant progress has been made over the last several years towards the development of algorithms for the acquisition and quantitative analysis of EMG signals to assist in the diagnosis and monitoring of neuromuscular disorders.
Routines that use individual MUPTs to estimate typical MUPs and calculate MUP feature values (such as duration, amplitude, area etc.) for each motor un it have been developed along with additional routines that, for each MUPT, extract macro MUPs using spike-triggered averaging, and calculate macro MUP feature values. Routines that use macro MUPs to obtain motor unit number estimate (MUNEs) and to study the size distribution of the MUs of a muscle have been developed. Routines to assess motor unit fibre density, MUP stability, and neuromuscular junction performance that use MUP acceleration to detect the significant contributions to detected MUPs of individual muscle fibres positioned close to the detection electrode are being developed. In addition, routines for robustly estimating motor unit firing behaviour, even for MUPTs that are incomplete or that contain errors have been developed. Finally, routines to graphically display and edit all of the various waveforms and histograms calculated following EMG signal decomposition have been developed.
Clinical Decision Support:
Current methods for decomposing composite EMG signals make it possible to easily collect large amounts of high quality QEMG data. However, it is still not known how to best quantitatively interpret and use the large volumes of data available. Therefore, techniques for extracting and interpreting important clinical information from available QEMG data need to be developed and evaluated. For these reasons, a comprehensive method for extracting patterns from mixed-mode training data for fuzzy classification and methods for incorporating the extracted patterns and fuzzy classification techniques into a transparent decision support system has been developed. These decision support methods need to be further developed and modified to facilitate interpretation of QEMG data.
EMG Signal Simulation:
An EMG signal simulation algorithm based on a physiologically realistic muscle model consisting of overlapping MUs with anatomically realistic muscle fibres with varying diameters and neuromuscular junction locations has been developed. MUPs are created by summing line-source based muscle fibre potentials calculated as detected by a needle electrode for each muscle fibre of a motor unit. EMG signals corresponding to varying amounts of disease can be created by modifying the muscle model to reflect the affects of myopathic and neuropathic disorders. Further development of the simulation techniques to more comprehensively model the effects of various diseases on muscles and the EMG signals they create is required.
Motor Unit Number Estimation:
Reliable estimates of the number of motor units comprising a specific muscle or group of small muscles supplied by a single nerve can be very useful in the diagnosis and management of neuromuscular disorders affecting alpha-motor neurons. Several different techniques have been developed for obtaining motor unit number estimates. These techniques are being refined and clinically evaluated. Motor unit number estimation involves the application and development of pattern recognition and statistical estimation techniques.
Decomposition-based Quantitative Electromyography (DQEMG):
The DQEMG methods described above are available on Compumedics Comperio DQ and Viasys Viking Select clinical EMG systems and are being used by several collaborating physicians. (Compumedics and Viasys are manufacturers of clinical EMG diagnostic equipment.)