This work propose a personalized protocol for muscle computer interface that uses finger movements as input commands. In previous studies, the same finger movements have been used for all subjects without considering personal characteristics. Finger movements that would be used as an input are different by individuals. As a first step, appropriate movements, which can be performed repeatedly and reliably, are selected based on the individual characteristics of multichannel surface electromyography (sEMG) on forearm and wrist. The average of number of selected movements is 22.8±8.7 for forearm and 17.6±6.1 for wrist by 40 healthy subjects among the every possible 62 finger movements. As a second step, we investigated the feasibility of the selected movements for pattern recognition. The averaged of classification accuracy was 81±7% for forearm and 89±5% for wrist. User could use more input commands using their finger movements by the proposed protocol considering the personal characteristics.
Related publications
1. Y Na, S Kim, S Jo, J Kim, Ranking hand movements for myoelectric pattern recognition considering forearm muscle structure, Medical & Biological Engineering & Computing, 55(8), 1507-1518, 2017. [LINK] [PDF]
2. Y Na, S Kim, S Min, C Choi, S Kim, S Jo, J Kim, Personalized protocol to select usable movements for myoelectric pattern recognition on forearm and wrist locations, MBEC 2014.[LINK][PDF]
