Key points are not available for this paper at this time.
With the ability to detect volumetric changes of contracting muscles, ultrasound (US) was a potential technique in the field of human-machine interface. Compared to the US imaging (B-mode US), the signal from a static single-element US transducer, A-mode US, was a more cost-effective and convenient way toward the real-world application, particularly the wearables. This study compared the performance of the single-channel A-mode US with single-channel surface electromyogram (sEMG) signals, one of the most popular signal modalities for wrist and finger gesture recognition. We demonstrated that A-mode US outperformed sEMG in six out of nine gestures recognition, while sEMG was superior to A-mode US on the detection of the rest state. We also demonstrated that, through feature space analysis, the advantage of A-mode US over sEMG for gesture recognition was due to its superior ability in detecting information from deep musculature. This study presented the clear complementary advantages between A-mode US and sEMG, indicating the possibility of fusing two signal modalities for the gesture recognition applications.
He et al. (2019) studied this question.