12954
0
Posted on May 28, 2018, 9 p.m.
New technology has been developed by researchers at North Carolina State University for decoding neuromuscular signals to control powered prosthetic wrists and hands.
Computer models are used that closely mimic behaviour of natural structures in the forearm, wrist, and hands; technology could be used to develop computer game interface devices for applications such as computer aided and gaming design. The newly developed technology has been working well in early testing and is awaiting clinical trials.
Current prosthetics rely on machine learning to create pattern recognition for prosthesis control, which requires user to teach the device to recognize specific patterns of muscle activity to be translated into commands such as opening and closing a prosthetic hand. Pattern recognition control requires a lengthy process of training, which can be tedious and time consuming.
Researchers have developed a user generic musculoskeletal model that is more intuitive, reliable, and more practical. Electromyography sensors placed on forearms of 6 able bodied volunteers tracked exactly which neuromuscular signals were being sent when performing various wrist and hand actions. Data collected was used to create the generic model which translated the neuromuscular signals into commands which manipulated a powered prosthetic.
When a hand is lost, the brain network is still wired as if the hand were still there. If a person wants to pick something up, the brain will still send the required signals. Sensors are used to pick up those signals and convey the data to a computer where it is fed into a virtual musculoskeletal model taking the place of muscle, joints, and bones calculating movements which would have happened if the hand and wrist were still whole, to perform the event in a coordinated way in real time resembling natural fluid movement motions.
Incorporating knowledge of biological processes behind generation of movement the team was able to create a novel neural interface for prosthetic which is generic and reliable across different arm postures, potential applications are not limited to prosthetic devices.
The team currently is seeking volunteers with transradial amputations to help with further testing and development of activities from daily life. The technology is not yet ready for commercial availability, but the model is compatible with available prosthetic devices. Incorporation of machine learning into the generic musculoskeletal model is also being explored, to allow user to gain more nuanced control and better adaptability to specific users in the long term.
Materials provided by North Carolina State University.
Note: Content may be edited for style and length.
Journal Reference:
Lizhi Pan, Dustin L. Crouch, He Helen Huang. Myoelectric Control Based on A Generic Musculoskeletal Model: Towards A Multi-User Neural-Machine Interface. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2018; 1 DOI: 10.1109/TNSRE.2018.2838448
