Automatic analysis and classification of surface electromyography.

In this paper, parametric modeling of surface electromyography (EMG) algorithms that facilitates automatic SEMG feature extraction and artificial neural networks (ANN) are combined for providing an integrated system for the automatic analysis and diagnosis of myopathic disorders. Three paradigms of ANN were investigated: the multilayer backpropagation algorithm, the self-organizing feature map algorithm and a probabilistic neural network model. The performance of the three classifiers was compared with that of the old Fisher linear discriminant (FLD) classifiers. The results have shown that the three ANN models give higher performance. The percentage of correct classification reaches 90%. Poorer diagnostic performance was obtained from the FLD classifier. The system presented here indicates that surface EMG, when properly processed, can be used to provide the physician with a diagnostic assist device.

Waveform estimation techniques for event-related bioelectric signals: a study of performance.

Many bioelectric signals result from the electrical response of a physiological system to an impulse that can be internal (ECG signals) or external (evoked potentials). A comparative study of performance of seven waveform estimation techniques used for event-related signals that are time-locked to a stimulus is presented in this paper. Computer generate 1 signals and noise for several signal-to-noise ratios (SNRs) are used to make ensembles of simulated noisy waveforms. The performance of each technique is numerically investigated using the root-mean-squared error and two well known SNR estimators. The results show that an adaptive impulse correlated filter performs the best. It is capable of estimating the deterministic component of the signal and removes the noise uncorrelated with stimulus even if this noise is colored and without the need for prealignment.

A pattern-analysis methodology for esophageal motility records.

The features of the esophageal contractile activity recorded during swallowing have been investigated using a signal-processing approach. Data of a 10 min recording taken from 10 normal subjects have been examined. The final features of each peristaltic wave are a set of parameters comprising the locations and magnitudes of the local wavelets contributing to the peristaltic. The approach is based on an inverse filtering technique. The inverse filter, designed with a knowledge of an average peristaltic profile formed by coherent averaging of many peristaltic examples aligned to their maxima, is used to refine the process of locating the local wavelets. It is shown that the inferred wavelet structure offers a good basis for modelling the different peristaltics, and that it suggests new parameters for esophageal characterization and classification procedures.

A fuzzy pattern recognition method to classify esophageal motility records.

Analysis and classification of esophageal motility records were investigated using signal processing and fuzzy-set pattern recognition techniques. A set of parameters has been extracted from the raw records and has previously been used as characterizing features. Improvements to these procedures were obtained by extracting these features from processed data, and some additional parameters were developed. The new set of features was used in the design of a fuzzy classifier, and classification accuracy was estimated using the leave-one-out technique. To our knowledge, this is the first report of the application of automatic classification to esophageal motility records.

Tauberian-prony feature extraction technique for esophageal motility patterns.

For the esophageal contractile activity recorded during swallowing, a feature extraction scheme has been developed. It recognizes the time, duration, and amplitudes of local peaks for each peristaltic wave. The method is based on the Tauberian approximation for modeling waveforms as a sum of identically shaped pulses with different time delays and amplitudes. Initial conditions on the pulse properties are set and an optimal solution is sought. The method is completely automated and can be utilized for characterization and classification purposes.