On the use of high-order cumulant and bispectrum for muscular-activity detection. (April 2015)

• Record Type: Journal Article
• Authors: Orosco, Eugenio; Diez, Pablo; Laciar, Eric; Mut, Vicente; Soria, Carlos; di Sciascio, Fernando
• Is Part Of: Biomedical signal processing and control. Volume 18(2015)
• Page Start: 325
• Page End: 333
• Publication Date: 2015-04
• Subjects: EMG -- HOS -- Cumulants -- Bispectrum; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2015.02.011
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: We propose novel third-order cumulant-based features for EMG signals. Classical bispectrum-based features were applied to EMG signals. A comparison between bispectrum-based features and cumulant-based features is proposed. Hypothesis: similar results can be achieved by bispectrum- or cumulant-based features. Analysis of the classification rates, Pearson's correlation, coefficient of determination and Friedman's test argued the principal hypothesis of this research. Abstract: The electromyographic (EMG) signals are extensively used on feature extraction methods for movement classification purposes. High-order statistics (HOS) is being employed increasingly in myoelectric research. HOS techniques could be represented in the frequency domain (high-order spectra, e.g., bispectrum, trispectrum) or in the time domain (higher-order cumulants). More calculus is required for computing the HOS in the frequency domain. On the one hand, classical bispectrum-based features were applied to EMG signals. We propose novel third-order cumulant-based features for EMG signals. Three different classifiers are implemented for muscular-activity detection. Different analysis and evaluations were applied to both HOS-based features in order to qualify and quantify similarities. Based on these results, it is possible to conclude that cumulant-based features and bispectrum-based features had comparable behavior and allowed similar classification rates. Hence, extra calculus in order to convert time- to frequency-domain should be avoided.