N.B. Jones

New signal processing techniques for the decomposition of EMG signals

This paper relates to the use of knowledge-based signal processing techniques in the decomposition of EMG signals. The aim of the research is to automatically decompose EMG signals recorded at force levels up to 20 per cent maximum voluntary contraction (MVC) into their constitutent motor unit action potentials (MUAPS), and to display the MUAP shapes and firing times for the clinician. This requires the classification of nonoverlapping MUAPs and superimposed waveforms formed from overlapping MUAPs in the signal. Nonoverlapping MUAPs are classified using a statistical pattern-recognition method. The decomposition of superimposed waveforms uses a combination of procedural and knowledge-based methods. The decomposition method was tested on real and simulated EMG data recorded at force levels up to 20 per cent MVC. The different EMG signals contained up to six motor units (MUs). The new decomposition program classifies the total number of MUAP firings in an EMG signal with an accuracy always greater than 95 per cent. The decomposition program takes about 15 s to classify all nonoverlapping MUAPs in EMG signal of length 1·0 s and, on average, an extra 9s to classify each superimposed waveform.

Selective noninvasive electrode to study myoelectric signals

The paper describes the design and construction of a selective surface electrode for use in a clinical environment. The main criterion of the design was to enable the recognition of individual motor unit action potential trains (MUAPTs) at moderate force levels. The main features of the electrode are, first, a small concentric bipolar arrangement to avoid electrode/muscle fibre alignment problems and to allow measurements within a small, well defined probed volume; secondly, the non-requirements for conducting paste or gel; and thirdly, the casing acting as an earth plate. All of these simplify its use. The results of tests undertaken with the electrode showed that it was able to pick up individual MUAPTs at up to 20 per cent of maximum voluntary contraction from the first dorsal interroseous muscle. Tests were carried out on the small hand muscles to further demonstrate the usefulness of the electrode. A computer program was written to calculate the shift in frequency of the power spectrum of the recorded myoelectric signal with muscle fatigue and hence indirectly to demonstrate the ability of the electrode to detect the reduction in muscle fibre conduction velocity.

A computer based system for the study of QT intervals

The authors have developed a system which scans electrocardiogram (ECG) waveforms stored on paper and converts them to digital data stored on a computer. The system enables quick and reliable measurements of QT intervals, thereby replacing the tedious and potentially insensitive method of hand measurements. Preliminary results of comparison between hand and user-interactive measurements are presented to show the accuracy and characteristics of the system.<<ETX>>

Optimised method for locating reference markers in rontgen stereophotogrammetric analysis of orthopaedic radiographs.

REFERENCE MARKERS implanted in the skeleton are essential for rrntgen stereophotogrammetric analysis (RSA) of orthopaedic radiographs (SELVIK, 1983). Very small movements between these reference markers or between them and an implant can be /neasured using RSA. Such technique has found applications in many areas of orthopaedics where motion analysis is important, such as migration of knee and hip prostheses, joint stability, joint kinematics and skeletal growth (BROSTROM et al., 1989; MJOBRG et al., 1986; RYD, 1992; SELVhK, 1989; BYLANDER et al., 1981). A conventional RSA investigation consists of four steps. Reference markers are first introduced into the bone using a stainless-steel cannula with a bevelled tip in combination with a hand-operated piston. In the second step, a pair of radiographic films are exposed. After obtaining the two radiographs, reference markers are then identified visually and their co-ordinates are measured manually using a Hasselblad high-precision measurement table. These coordinates are used in the final step by computer programs to calculate the relative positions and motions. The measurement accuracy of RSA is estimated to be within 10--250 pan and 0.03-0.6 ~ (KARRHOLM, 1989). Conventional RSA is a very accurate method as well as a complex operation that requires a calibration cage, a Hasselblad digitising table, special radiographs and software for data analysis. A simplified RSA scheme has been developed, which uses the standard anterior-posterior and medio-lateral radiographs and does not need a calibration cage, a Hasselblad digitising table and special X-ray tubes (WALKER and SAHASIVAM, 1992). The simplified RSA still requires a high-precision digital tablet and manual operation to locate reference markers, which are usually made of spherical tantalum balls and show as bright beads on radiographs. The number of markers used depends on the object being studied and the accuracy required, but for a complete kinematic analysis at least three noncollinear markers must be implanted. Usually, several repeated readings for each reference marker are made to improve the accuracy and reproducibility of manual operation. Although manual measurement of the co-ordinates of the reference markers can achieve a reasonable accuracy of around 4-0.05 ram, it requires special and costly equipment and is a tedious and time-consuming task (WALKER and SAHASIVAM, 1992). If this part of RSA operation is improved, wider use could be made of it. Recent advances in personal computer and scanner technology have resulted in popular and cheap methods for digitising radiographs and image processing. Nowadays a PC-portable image digitiser can readily achieve a digitising precision of 0.05 mm. In this work, an alternative method is presented that can automatically locate reference markers and measure their co-ordinates using a PC-portable image digitiser.

Intelligent classification in EMG decomposition

This paper presents research relating to the use of computers for the intelligent decomposition of myoelectric signals (EMG). A knowledge based expert system is described which decomposes superimposed waveforms formed from overlapping motor unit action potentials (MUAPs) in a myoelectric signal using symbolic information provided by numerical recognition analysis. The system, written in Prolog, consists of some 30 rules in the knowledge base that are driven by an interpreter that incorporates uncertain reasoning based on fuzzy set theory. The expert system contains both procedural and declarative knowledge representations of the problem domain. The declarative rules contain a description of the relationships between the raw motor unit (MU) information collected by the numerical analysis and the superimposed waveforms being decomposed. The procedural rules interact with the declarative rules through rule attachments that activate demon procedures. The demon procedure computes fuzzy certainty factors for all the possible combinations of MUAPs that form a superimposed waveform.