Use of high-density surface-electromyography of the forearm as a method for predicting thumb rotation in myoelectric transradial prosthetics

• Main Author: Aranceta-Garza, Alejandra
• Published: University of Strathclyde 2015
• Subjects: 610.28
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.665240

Until now, natural thumb control mechanisms are lacking in the upper limb prosthetic development. This lag is due to the complex anatomy of the musculature of the hand making the upper limb prosthetic research a very complicated area. Several studies have attempted to better our understanding of the neural control of the hand. With applications including clinical rehabilitation, surface-Electromyography (sEMG) has been steadily improving the knowledge in this area, albeit still very limited. The development of high-density sEMG (HD-sEMG) has drastically increased the sensitivity of EMG techniques. Despite this research effort, there are significant gaps in the field. Furthermore, current data analysis is almost exclusively performed off-line and so, neurally controlled prostheses are limited to research labs and are not a clinically viable technology. Therefore, it is evident that new technologies are required to understand the dexterity of the human hand for prosthetic control. A common theme across the different hand-prosthetic developers is not to have mechanisms to drive the thumb based on muscular contractions. Due to the lack of intuitiveness for an amputee to operate the prosthetic device, it requires several highly demanding training sessions between the patient and the prosthetist. These sessions are oriented for the amputee to be able to control in duration and magnitude, the contractions of the chosen muscles to drive the motors of the prosthesis. In this research, the muscle activity from the forearm is identified and correlated with specific hand movements leading to improve the commercially available myoelectric transradial prosthetics. This was achieved through the understanding of sEMG patterns related to differentiation of thumb op position to different fingers. The acquired signals were investigated based on time-domain analyses (i.e. amplitude signal analysis, root-mean square values, statistical analyses), followed by a joint time and frequency-domain analyses (i.e. coherence estimate and cumulant analysis). Finally, unsupervised machine learning techniques were applied aiming to differentiate the different sEMG patterns during the different thumb opposition. This differentiation leads to a better understanding with regards to prospective controller mechanisms aiming to develop new prosthetic devices enhancing the experience of transradial amputees with the use of their prosthetic.