Éric Le Carpentier

Arthur Bureau, Jean-Maxime Le Carpentier, Eric Le Carpentier et al.

The electromyogram (EMG) in needle detection represents one of the steps of the electroneuromyogram (ENMG), an examination commonly performed in neurology. By inserting a needle into a muscle and studying the contraction during effort, the EMG provides extremely useful information on the functioning of the neuromuscular system of an individual, but it is an examination that remains complex to interpret. The objective of this work is to participate in the design and evaluation of a software allowing an automated analysis of EMG tracings of patients suspected of neuromuscular diseases, orienting the diagnosis towards either a neuropathic or myopathic process from recorded tracings. The software uses a method of signal decomposition according to a Markovian model, based on the analysis of motor unit potentials obtained by EMG, then a classification of the tracings. The tracings of 9 patients were thus analyzed and classified on the basis of the clinical interpretation of the neurologist, making it possible to initiate a "machine learning" process. The software will then be submitted to new tracings in order to test it against a practitioner experienced in EMG analysis.Translated with www.DeepL.com/Translator (free version)

D. Ge, J. Idier, E. Le Carpentier

This paper proposes and compares two new sampling schemes for sparse deconvolution using a Bernoulli-Gaussian model. To tackle such a deconvolution problem in a blind and unsupervised context, the Markov Chain Monte Carlo (MCMC) framework is usually adopted, and the chosen sampling scheme is most often the Gibbs sampler. However, such a sampling scheme fails to explore the state space efficiently. Our first alternative, the $K$-tuple Gibbs sampler, is simply a grouped Gibbs sampler. The second one, called partially marginalized sampler, is obtained by integrating the Gaussian amplitudes out of the target distribution. While the mathematical validity of the first scheme is obvious as a particular instance of the Gibbs sampler, a more detailed analysis is provided to prove the validity of the second scheme. For both methods, optimized implementations are proposed in terms of computation and storage cost. Finally, simulation results validate both schemes as more efficient in terms of convergence time compared with the plain Gibbs sampler. Benchmark sequence simulations show that the partially marginalized sampler takes fewer iterations to converge than the $K$-tuple Gibbs sampler. However, its computation load per iteration grows almost quadratically with respect to the data length, while it only grows linearly for the $K$-tuple Gibbs sampler.