Browsing by Author "Taffler, Sean"

Now showing 1 - 3 of 3

Open Access
A STUDY OF EMG SIGNALS FROM LIMBS WITH CONGENITAL ABSENCE AND ACQUIRED LOSSES
(2002) Kyberd, Peter; Taffler, Sean
The basic assumption made by the designers of EMG amplifier systems is that the substantial difference between different person's detected EMG signals is the energy generated, but the spectral content is sufficiently similar to be ignored. Anecdotally, there is little or no expectation that there is any structural difference between the muscles of persons with congenital absence and those with an amputation, although newer evidence suggests this is untrue [1]. This approximation works well for simple amplifier/detector systems that were common a decade past [2], but with the increasing availability of compact on-line processing devices and the increase in differing EMG processing schemes this approximation is likely to prove unreliable [3]. This study was made with persons who attended the Oxford Limb Fitting Centre at the Nuffield Orthopaedic Centre, Hospital in Oxford UK, from 1989 to 2001 as part of a program looking at novel means of processing, and the analysis of EMGs [4,5]. It sought to understand the variability of the results as they were derived from the techniques applied (Fuzzy logic, Neural Networks and Wavelet transforms). While undertaking the study the new information from the structural studies in Texas [1], showed that it was a worthwhile exercise, and one likely to produce unexpected results.
Open Access
The Use of Fuzzy Logic In the Processing Of Myoelectric Signals
(1999) Taffler, Sean; Kyberd, Peter J.
This paper describes the use of Fuzzy logic for the processing of EMG signals. This can increase the recognition rate and significantly reduce the number of computations required to generate an output. The initial placement of the Fuzzy sets was accomplished with the use of neural network techniques, these are not required for in the final system, only for setting up. The effectiveness of the features extracted from the EMG signals has been assessed using Principal Component Analysis (PCA) The developed system exhibits good generalisabilty but performs better when tuned to the intended user.
Open Access
The use of the Hilbert transforra in EMG Analysis
(1999) Taffler, Sean; Kyberd, Peter J.
The Fourier transform has traditionally been used for the detailed analysis of EMG signals. This has yielded many useful results, none more so than the descriptions of the energy produced at differing frequencies. This has been invaluable in the development of robust EMG controllers and the analysis of active or diseased muscle. Recently, Wavelet analysis has been applied to the study of EMG signals and it has provided additional insight into the underlying structure of the signal. Both these methods have drawbacks, the Fourier transform relies on analysis of complete wavelengths to describe a signal Wavelet analysis cannot resolve any event less than the length of the fundamental Wavelet. These factors manifest themselves as a smudging or broadening of the spectrum and therefore they lead to inprecissions in the results. Empirical Mode Decomposition (EMD) and the Hilbert Transform (HT) have been applied to analyse the the EMG signal. This is a method that is used extensively in the fields of seismology and meteorology and is now being applied to biological data. It is particularly good a resolving signals that are not based on continuous sinusoids. It has been used on EMGs to show that the energy in the signal is significant at frequencies up to 2KHz. The paper will present the results of a study of signals derived from a range of prosthesis users and non-users. The results from the Hilbert transform will be compared with results obtained using conventional methods of analysis.