Wavelet and Unsupervised Learning Techniques for Experimental Biomedical Data Processing

Wavelet and Unsupervised Learning Techniques for Experimental Biomedical Data Processing

Matteo Cacciola (Department of Civil Engineering, Energy, Environment and Materials, “Mediterranea” University of Reggio Calabria, Reggio Calabria, Italy), Salvatore Calcagno (Department of Civil Engineering, Energy, Environment and Materials, “Mediterranea” University of Reggio Calabria, Reggio Calabria, Italy), Filippo Laganà (Department of Civil Engineering, Energy, Environment and Materials, “Mediterranea” University of Reggio Calabria, Reggio Calabria, Italy), Francesco Carlo Morabito (Department of Civil Engineering, Energy, Environment and Materials, “Mediterranea” University of Reggio Calabria, Reggio Calabria, Italy), Diego Pellicanò (Department of Civil Engineering, Energy, Environment and Materials, “Mediterranea” University of Reggio Calabria, Reggio Calabria, Italy), Isabella Palamara (Department of Civil Engineering, Energy, Environment and Materials, “Mediterranea” University of Reggio Calabria, Reggio Calabria, Italy), Mario Versaci (Department of Civil Engineering, Energy, Environment and Materials, “Mediterranea” University of Reggio Calabria, Reggio Calabria, Italy) and Biagio Cammaroto (Department of Civil Engineering, University of Messina, Messina, Italy)
DOI: 10.4018/ijmtie.2012070105
OnDemand PDF Download:
$37.50

Abstract

Learning theories and algorithms for both supervised and unsupervised Neural Networks (NNs) have already been accepted as relevant tools to cope with difficult problems based on the processing of experimental electromagnetic data. These kinds of problems are typically formulated as inverse problems. In this paper, in particular, the electrical signals under investigations derive from experimental electromyogram interference patterns measured on human subjects by means of non-invasive sensors (ElectroMyoGraphic, sEMS surface data). The monitoring and the analysis of dynamic sEMG data reveal important information on muscles activity and can be used by clinicians for both preventing dramatic illness evolution and improving athletes performances. The paper proposes the use of the Independent Component Analysis (ICA), an unsupervised learning technique, in order to process raw sEMG data by reducing the typical “cross-talk” effect on the electric interference pattern measured by the surface sensors. The ICA is implemented by means of a multi-layer NN scheme. Since the IC extraction is based on the assumption of stationarity of the involved sEMG recording, which is often inappropriate in the case of biomedical data, we also propose a technique for dealing with non-stationary recordings. The basic tool is the wavelet (time-frequency) decomposition, that allows us to detect and analyze time-varying signals. An auto-associative NN that exploits wavelet coefficients as an input vector is also used as simple detector of non-stationarity based on a measure of reconstruction error. The proposed approach not only yields encouraging results to the problem at hand, but suggests a general approach to solve similar relevant problems in some other experimental electromagnetics applications.
Article Preview

1. Introduction

Most relevant medical problems are today faced by processing (by visual inspection or some automatic means) electrical signals detected on the human body. The evaluation of patient populations often includes the use of ancillary tests for diagnosis and/or prognosis. Data sets collected from these diagnostic tests, such as the Electroencephalogram (EEG), the Electromyogram (EMG), the Electrocardiogram (ECG) and, more recently, the functional Magnetic Resonance Imaging (fMRI), tend to be complex, large and high-dimensional. The trend towards the digitization of the traditionally analog EEG, EMG and ECG signals has coincided with the development of computing power and multivariate signal processing techniques capable of manipulating and analyzing such large data sets (Akay, 1997).

The use of Independent Component Analysis (ICA), an unsupervised learning technique which generalizes the Principal Component Analysis (PCA), commonly implemented through Neural Network (NN) schemes, is proposed in this study to process experimental biomedical data. Applied to sEMG (ElectroMyoGraphy surface) data, ICA results in numerous spatially- independent patterns, each associated with a unique time-course, providing a way to separate different electrical signals coming from different muscle activities (Junk et al., 2000). In contrast to the variable nature of the EMG surface recorded from a single muscle in isolation, the ICA of the sEMG from several muscles simultaneously allows the detection of highly reproducible components for example in the sEMG of the face and the throat during swallowing, and in the sEMG of arm muscles during the movements reaching (McKeown, Torpey, & Gehm, 2002).

The researchers in the present work show important applications in the study of some neurological diseases, and in the monitoring of athletic activities to improve significantly the potential of the athletes as well as the capabilities of normal subjects in daily actions, since it makes it possible, in principle, to enhance motor coordination. Also, musculo-skeletal disorders are the first cause of patient-physician encounters in the industrialized countries (McKeown, 2001). This paper is organized as follows. In Section 2 the type of data coming from electrical activity of muscles will be discussed. In Section 3 we shall propose the McKeown’s idea of motion through integration of sub-movements (McKeown, 2000). The computational model incorporating sub- movements will be presented in Section 4. Section 5 is devoted to the proposal of NN schemes to implement ICA. Section 6 will report the results achieved by processing the experimental data. The assumption of stationarity of the electrical signals will be relaxed in Section 7, where the wavelet approach will be proposed. Finally, some conclusions are drawn.

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 6: 2 Issues (2017): Forthcoming, Available for Pre-Order
Volume 5: 2 Issues (2015)
Volume 4: 2 Issues (2014)
Volume 3: 4 Issues (2013)
Volume 2: 4 Issues (2012)
Volume 1: 4 Issues (2011)
View Complete Journal Contents Listing