ABSTRACT
In this work, the problem of movement artifacts, affecting dynamic myoelectric recordings, is studied. Dynamic myoelectric recordings allow the study of muscle activation intervals which provide a fundamental information for both the movement analysis and the clinical assessment of motion disorders. The detection of muscle activation patterns can be detrimentally affected by the presence of artifacts due to the movement of the surface electrodes on the skin. In order to overcome this difficulty, three artifacts filtering procedures have been tested: a high-pass filtering procedure, a moving average procedure and a moving median procedure. For the monitoring of the correct detection of muscle activation intervals, before and after the artifacts removal, a new original double-threshold statistical detector, recently developed by one of the authors, is utilised. This detector shows good performance and it is not affected by subjective settings, such as the classical activation intervals detector based on a not automatical thresholding of the integrated myolectric signal. Preliminary results illustrate the effectivness of the artifacts removal. The moving median filtering procedure seems to perform better than the other two proposed methods.
