Feature Extraction and Selection for Myoelectric Control Based on Wearable EMG Sensors

• DOI: 10.3390/s18051615
• Publication Date: 2018-05-21T08:07:30Z
• Authors: Angkoon Phinyomark; Rami N. Khushaba; Erik Scheme
• Keywords: Wearable Technology

Specialized myoelectric sensors have been used in prosthetics for decades, but, with recent advancements wearable sensors, wireless communication and embedded technologies, electromyographic (EMG) armbands are now commercially available the general public. Due to physical, processing, cost constraints, however, these typically sample EMG signals at a lower frequency (e.g., 200 Hz Myo armband) than their clinical counterparts. It remains unclear whether existing feature extraction methods, which largely evolved based on sampled 1000 or above, still effective use emerging lower-bandwidth systems. In this study, effects of sampling rate (low: vs. high: Hz) classification hand finger movements were evaluated twenty-six different individual features eight sets multiple using variety datasets comprised both able-bodied amputee subjects. The results show that, average, accuracies drop significantly ( p.