A need exists to increase the functionality of myoelectric prostheses without increasing the mental requirement of operation. Implantable myoelectric sensors have made it possible to record multiple muscle activities with high fidelity. Given this high dimensionality of inputs, what is the best way of implementing control? Muscle synergies have been proposed for coordinating the many degrees-of-freedom (DOF) of the neuromotor system. This work aimed to investigate synergies as a viable control paradigm for multi-DOF myoelectric devices with regard to the properties of robustness, scalability, and volitional activation. First, this work investigated if muscle synergies formed a predictive basis set for muscle coordination patterns associated with a variety of hand postures. Subjects mimed hand postures of the American Sign Language alphabet while electromyographic (EMG) activity was recorded from hand muscles. Non-negative matrix factorization (NMF) showed that a small number of hand postures could establish a robust set of synergies for predicting the EMG patterns of a variety of hand postures. Second, this work investigated the scaling of muscle synergies in hand grasping at sub-maximal force levels. Subjects performed a force-tracking task using different grasps while EMG was recorded. Statistical and NMF analyses showed that the primary synergies of grasping retained their structures and scaled linearly with grasp force. Third, this work investigated, through a virtual target reaching task, the volitional control of multiple DOFs using muscle synergies versus single-muscle inputs. It was hypothesized that users could more intuitively achieve independent and simultaneous control of myoelectric inputs using muscle synergies over single-muscle activations. The results showed that while users were able to independently and simultaneously modulate synergy activations, this control paradigm was statistically no better than one based upon single-muscle inputs. From these investigations, it is concluded that while muscle synergies exhibit useful properties for control such as robustness, generalizability, and scaling, their practical benefit in a volitional control task is not significantly greater than a single-muscle control paradigm. Results from these investigations also suggest that the method of control implemented by the neuromotor system is not bound by muscle synergies, but rather by a combination of both synergy and single-muscle activations

Last modified

• 08/27/2018

Creator

• Abidemi Bolu Ajiboye

DOI

• https://doi.org/10.21985/N2XT6F

Subject

• Biomedical Engineering

Keyword

• muscle synergy
• prosthesis
• EMG
• myoelectric control
• dimensionality reduction
• hand

Date created

• 2007-12-04

Resource type

• Dissertation

Rights statement

• In Copyright