A Neural Interface for Artificial Limbs: Targeted Muscle Reinnervation

Providing adequate control of a powered artificial arm is difficult, especially with high levels of amputation where the need is greatest.  In this presentation we will describe a new technique to create a bi-directional neural interface for artificial limbs called ‘targeted muscle reinnervation’ (TMR).  With TMR it is possible to take the residual nerves to in an amputated limb and transfer them to spare muscle and skin in or near the limb.  The nerves grow into this muscle, and then the surface EMG over this muscle can be used as an additional control signal.  For example, if the median nerve reinnervates a small region of surface muscle, then when the amputee thinks ´close hand´ this muscle will contract and the myoelectric signal can be used to close the powered hand. Since physiologically appropriate neural pathways are used, the control is intuitive, thus easier and faster for the amputee. 

Similarly, sensory nerves can be transferred to the residual nerves so that skin of the chest or arm is reinnervated—targeted sensory reinnervation (TSR).  Then when the amputee is touched on this reinnervated skin, it feels like he or she is being touched in the missing arm or hand.  TSR can provide a pathway for true sensory feedback of light touch, graded pressure, sharp/dull, and thermal feedback. Research is presented showing how the skin of residual limbs has been reinnervated by hand afferents and our early attempts to provide closed loop feedback.

To date, over 40 people worldwide with shoulder disarticulation and transhumeral amputations have had successful targeted reinnervation surgery. Over 95% of the nerve transfers are successful in producing measurable EMG signals. With conventional prosthesis using simply the amplitude of the EMG for control, functional testing has shown marked improvement in all tests including a 50-300% increase in the blocks and box test, a 25-50% increase in speed with a clothes pin relocation tests and in ADL style testing such as the AMPS test.

Pattern Recognition computer algorithms have been added to greatly increase the degrees of freedom that can be intuitively operated.  Data are presented showing the speed and robustness of control. Videos are presented showing patients operating advanced motorized arm prostheses, demonstrating control of 3 functions at the shoulder, the elbow, a two function wrist and a multifunction hand capable of several different hand grasp patterns.  Patients can operate all of these degrees of freedom in an intuitive manner